Agent status at a glance
Blume watches agent activity in one narrow desktop sidecar so you can see whether an agent is running, finished, or waiting for approval.
Blume
Blume is a desktop sidecar for watching coding agents, reviewing their setup, and suggesting local-first improvements to rules and hooks. It helps users of tools like Cursor, Claude Code, and Codex understand what each agent is doing without leaving their machine.
What Blume does
Blume is a desktop sidecar for monitoring and improving coding agents. It is built for people who use tools such as Cursor, Claude Code, and Codex and need a clearer view of what each agent is doing, what files it touches, and which rules or skills are shaping its behavior.
The product combines agent monitoring, setup visibility, usage tracking, and on-device suggestions for rulebook cleanup. According to the site, it is designed local-first: conversation history stays on your device, and review workflows let you inspect evidence and diffs before you accept a suggestion.
Features
Unified view of agent setup
It brings together hidden files, skills, hooks, and rules across supported tools so the setup behind each agent is easier to inspect.
On-device improvement suggestions
Blume reviews rules, skills, and hooks on-device and proposes changes when it detects a mismatch between your instructions and the current setup.
Reviewable change workflow
Suggested changes include an evidence view and exact diff, with options to apply, dismiss, or save for later.
Usage tracking
The product shows usage across providers such as Claude, Codex, and Cursor so you can see plan limits and token burn before you hit a wall.
Local-first data handling
The privacy page says conversation history is stored locally on the device, and usage and spend stay local rather than being sent to Blume servers.
Common use cases
Track active agent work
Use Blume to keep a running view of whether each coding agent is active, completed, or waiting on you, instead of checking multiple harnesses manually.
Audit agent configuration
Use it to inspect hidden files, rules, hooks, and skills across supported tools when you want the agent setup in one place rather than scattered across repositories and providers.
Monitor provider usage
Use it when you want usage and spend visibility for Claude, Codex, or Cursor before plan limits or token burn interrupt work.
Approve or reject setup improvements
Use it to review suggested rulebook fixes, compare the evidence and diff, and decide whether to apply or dismiss the change.
Keep agent context local
Use it to keep conversations and processing local on your machine while still getting analysis features that help improve agent behavior.
Pros and Cons
Pros
- Shows live agent state and the rules, skills, and instructions behind that work.
- Keeps conversation history local on the device, according to the privacy page.
- Surfaces usage and spend locally so you can monitor provider limits before hitting a wall.
- Proposes rule, skill, or hook fixes on-device and lets you review the evidence before applying them.
- Combines setup visibility and improvement suggestions in one desktop companion.
Cons
- The provided sources do not include a working pricing page, so plan details are not visible here.
- Docs coverage is limited in the supplied material, so setup depth and full provider support are not fully documented.
FAQ
What is Blume for?
Blume is a desktop companion for people using coding agents such as Cursor, Claude Code, and Codex. It watches agent activity, surfaces setup information, and suggests improvements when chat instructions and project rules drift apart.
Is Blume local-first?
The source material describes Blume as a desktop app and sidecar for macOS workflows. It is designed to read conversation transcripts from supported AI tools on the device and store conversation history locally in a SQLite database.
Which coding agents does Blume support?
The docs mention setup guides for connecting coding agents and fixing usage tracking. The product page also says Blume works with Cursor, Claude Code, Codex, omp, and Pi, though the documentation provided here does not include full setup coverage for every provider.
How does Blume handle suggested improvements?
Blume can review rules, skills, and hooks on-device, then propose a fix for the user to approve, dismiss, or save for later. The product page also says it can review exact diffs before changes land.
What does Blume cost?
A pricing page was not available in the provided source set. The available pages do not show a free tier, trial, or paid-plan structure.
Quick Facts
- Category
- Developer Tool
- Platform
- Desktop app
- Primary use
- Monitor and improve coding agents
- Source domain
- blume.codes
- Supported tools mentioned
- Cursor, Claude Code, Codex, omp, Pi
- Data handling
- Local-first storage with optional cloud features for some improvements
Alternative a Blume
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.
Manta AI
Manta AI is an autonomous web app testing tool for teams that want to map application behavior, catch regressions, and generate tests without writing scripts or maintaining selectors. It works from a URL and supports plain-English test flows, run results with screenshots, and scheduled or deployment-triggered checks.
SonOf
SonOf connects to your repo and PM tool, audits the codebase and surrounding product context, and turns approved work into shipped tickets with senior engineering review. It is aimed at founders and engineering leaders who need backlog help without hiring a full team immediately.
Ghost
Ghost è un assistente AI da terminale per chattare, generare codice ed eseguire task da riga di comando. Include modelli gratuiti, supporta Linux, macOS e Windows, ed è open source.